Quote:
Originally Posted by amg6975
I own a company that makes such things and have been thinking about this for a while. The hard part is the analysis/configuration software, all the good packages have proprietary data formats so you can't just use their software unless you're using their hardware. The hardware/firmware is fairly trivial.
A simple DIY thing to get analog data onto the CAN so an AiM can record it is a good stop-gap for a project like this.
|
your the right man to speak to then!
what fe1x is doing and what he could do further down the track imo only requires raw data dumped into a .csv that is timestamped and labelled CH1, CH2, CH3.... etc. most generic data loggers i've found do not have high sampling rates, enough inputs, or outrageously expensive.
then it is excel, a graphics package or fit for purpose analyses software. at some point excel is too slow with a lot of data that you might get with high data frequency, if you were logging multiple channels at 200Hz around a race track. but that's where the graphics packages (particularly ones with math functions) come in.
i'm not familiar with the compatibility issues on other brands - but as an example syvecs "SView" (which is freely downloaded) will graph both the ECU data logged file and import anything from a .csv. so both ECU data and imported generic data can be assessed together. i've imported my old jb4 and mhd data straight into SView no problem. you can also create your own math functions using imported data, that could be as simple as conversions for voltages to more complex analyses.
although it would be nice if you could directly program the analogue voltages to actual units of interest within the data logger, its not really required as it can be done post logging.
third party specialist chassis software should also import .csv raw data since they would generally not be tied to one of the hardware manufactures.
if the end user needs compatibility then that is not the market i am thinking of. realistically no analyses is done in the drivers seat so getting this data into a dash logger isn't required. the basics for chassis work;
-Steering angle
-Throttle input
-Damper pots
-Accurate accelerometer.
-Speed
8x channels.
but someone like fe1x could fill up another dozen thermocouple inputs just 'cause....
while i have your attention, what would be the best currently available solution to program inputs to generate an output?
so the situation is I have a standalone ecu, but i dont have many inputs left. in the future i'd like to add 6 EGT thermocouples, ideally i would like to write code that would assess the EGT data for abnormalities. i'm thinking like using a rolling averages per cylinder, and for all 6 to establish what is normal in real time then monitoring individual cylinders against the established "normal". If one cylinder spikes in either direction (rate of change and amplitude comparison), then code an output channel to 5v instead of 0v - feedback that single input into my ecu and use it as a limp mode trigger.
too complex? googling i found this, general purpose input / output board:
http://megasquirt.info/products/diy-kits/gpio/